Print

Print


Hallo conlangers!

On Wednesday 30 November 2011 05:43:29 Gary Shannon wrote:

> It occurred to me to play around with "compiling" human language into
> some lower level language in a manner analogous to the way a
> high-level computer language is compiled into a low level language
> like assembler. For example, a C++ statement like:
> 
> a = b + c;
> 
> might become something similar to the hypothetical:
> 
>  INT a // Declare integer variable a and reserve memory space for it
>  INT b // Declare integer variable b and reserve memory space for it
>  INT c // Declare integer variable c and reserve memory space for it
> ...
>  LDA b // load value of variable b into the accumulator
>  ADD c // add the value of variable c
>  STA a // store the result into the variable a
> 
> Where each high level instruction becomes some number of simpler
> primitive "machine level" instructions.
> 
> Now consider a natlang statement like:
> 
> "John gave the book of poetry to Mary."
> 
> There is a whole lot going on here that needs to be spelled out
> explicitly in the "machine level" compilation. First, we have to
> recognize that "John" and "Mary" are "variables" declared as instances
> of the general class "Person" (as opposed to "integer" or "string" as
> one might find in a computer language) and that "the book" is a
> variable declared as an instance of the class "Book". We can give them
> any arbitrary name, and explicitly assign to them whatever values
> (names, identities, characteristics, etc...) we are given.
> 
> Declarations take the form "Type Variable_name" and statements take
> the form S V S where "S" is any substantive (i.e., variable or
> constant) and "V" is any verb-like operator that expresses a
> relationship between the two substantives.
> 
>  Person W // declare variable "W" to be of type "Person"
>  Person X // declare variable "X" to be of type "Person"
>  Book Y // declare variable "Y" to be of the type "Book"
>  Subject Z // declare variable "Z" to be of the type "Subject"
>  W name_is "Mary"
>  X name_is "John"
>  Z name_is "poetry"
>  Y is_about Z
>  X gave Y
>  W accepted Y

This somehow reminds me of a dormant project of mine, called X-1.
What little has come out of it so far can be seen here:

http://www.frathwiki.com/X-1

I haven't done anything on that for quite a while and marked
it abandoned, though I am considering reviving it as an
interstellar lingua franca in a science fiction setting I am
collecting ideas for.
 
> Now the question becomes, given statements in different languages that
> express essentially the same meanings, if those statements are broken
> down into "assembly language" will they be the same, or at least very
> similar, regardless of the original language of the statement?
> 
> In other words, could such a conlang (that I called here "assembly
> language") really be the "universal machine language" of all human
> communication? (No I'm not talking about Chomsky's universal grammar
> because the grammar of the source language has nothing to do with the
> grammar of "assembly language". If anything it tries to capture a
> universal semantics.) And what "V" operators would be needed to
> implement such a language?

I don't think human language processing is much like programming
language processing in a computer (at any rate, human languages
are obviously "interpreted" rather than "compiled"), and the notion
of a "universal machine language" of human communication seems
meaningless to me.

--
... brought to you by the Weeping Elf
http://www.joerg-rhiemeier.de/Conlang/index.html
"BÍsel asa  m, a  m atha cvanthal a cvanth atha  mel." - SiM 1:1